Long term data preservation by 3D printed stamped clay

As a 2020 lockdown project, we try to find an innovative solution to protect data (text or binaries as neural networks) for a long term period.

The issue

The data preservation is a difficult problem including material science, data science and cognitive science as the data has to be, after eons, in good shape, readable and understandable. The Arctic world Archive has a special storage in Spitzberg. The data are stored on microfilms. The preservation time should be 500 to 1000 years. But how we could store information for 10000, 20000 years.

A Jugaad approach

In a frugal innovation approach, we consider what are the most ancient readable data. We can find seven:

The palermo stone (4300 years) – The Reform of Urukagina ( 4500 years) in stone, The instructions of Shuruppak (4600 years) this is a tablet, The Kesh temple Hymn (4600 years) a tablet also, Proto elamite tablets (5200 years), Narmer Palet (5200 years) in stone and the Kish Tablet (5500 years)

Most of them are thus in clay and represent text as a collection of symbols acting together as a rebus.

So we developed a technique to produce modern clay tablets using a combination of CAD and 3D printing.

Conversion to a cuneiform tool

In order to print text on wet clay, we need a tool that allow to imprint the material avoiding bubbles and errors.

The text will be splitted in line, and the binaries (images, neural networks…) converted in base 85.

You can test the generator at https://cunei.herokuapp.com . This is a demo on a free instance so it takes some time to start.

Interface of the tool builder

The tool builder makes a 3D printable STL file of the text or a base85 converted file. The toll is mirrored and chars are beveled in order to allow a better penetration in wet clay.

3D Printing and imprinting

The STL file can be printed on our 3D45 Dremel 3D printer. In order to get a good solidity, the material will be eco-abs 80% filled with a precision of 0.1mm.

The ABS tool can now be pressed on a wet clay tablet.

First attempt of imprinting

The resolution as well as the amount of data that can be stored depends on the quality of the material and its viscosity. Clays may a non-newtonian bingham fluids so to get a better result the pression has to be allied very slowly (and the tool can be remove quickly)

You can find two type of clays. The usual ones has to be post processed in a furnace at 1300°C for hours but you can also find clays that drie at room temperature within 5 days.

How many time we may hope?

As usual, the time the artifacts will be readable is pure speculation. It’s also depends on the amount of energy and temperature will occur during the apocalypse;-) Nevertheless, if we take archeological facts as a reference, clays have a better resistance to heat than paper or microfilm and don’t require to be stored in Arctic. The above mentioned 5000 years old tablets were not made for long term data preservation and they are usually found in ruins and not especially protected. Furthermore, our knowledge in materials allow to think about new formulation of clays that could be extremely stable. One of the main issue could be the mechanical fragility but it is possible to mix clays with cellulose in order to improve their resistance.

If the lockdown lasts… I’ll propose you something for large neural networks storage and also on clays material formulation.

The source code is available at GitHub

Télécharger cet article au format PDF ou ePub

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.